A road condition judgment method and system for large-scale logistics transportation

Abstract

The present invention proposes a road condition judgment method and system for large-scale logistics transportation. It relates to the application field of intelligent transportation systems, wherein the method includes: obtaining all accessible paths from the departure point to the destination of large-scale logistics transportation, obtaining the road condition time series data of each accessible path in the historical and current time periods; pre-training a road condition prediction model based on the road condition time series data of the historical time period, inputting the road condition time series data of the current time period into the road condition prediction model, obtaining the predicted road condition time series data of each accessible path in the set time period, and selecting the optimal path; when driving along the optimal path, obtaining the adjacent route segment of the next route, predicting the traffic volume of the next route, and if the traffic volume of the next route differs from the predicted traffic volume by more than a preset threshold, finding a replacement route for the next route. Through this method and the corresponding system, the route can be dynamically predicted and adjusted to cope with sudden changes in road conditions.

Description

Technical Field

[0001] The present invention proposes a road condition judgment method and system for large-scale logistics transportation, and relates to the application field of intelligent transportation systems. Background Art

[0002] In the logistics and transportation industry, the quality of road conditions directly affects the efficiency and cost of cargo transportation, especially when it comes to long-distance transportation across cities or internationally. Good road conditions can ensure that goods arrive at their destination quickly and safely, while poor road conditions may cause delays, increase fuel consumption, and even damage goods. Therefore, judging road conditions in transportation plans has become a key link in logistics management.

[0003] Existing road condition judgment technologies mostly focus on describing current or past conditions, but have weak prediction capabilities for future road conditions, which limits their application in planning and adjusting logistics and transportation plans. In addition, existing technologies, based on road condition information collected by traditional data acquisition equipment, have relatively weak capabilities for analyzing road conditions and estimating traffic evolution trends on various sections of the traffic network. Summary of the invention

[0004] The present invention provides a method and system for determining road conditions for large-scale logistics transportation to solve the above-mentioned problems:

[0005] The present invention proposes a method for determining road conditions for large-scale logistics transportation, the method comprising:

[0006] Obtain all accessible routes from the departure point to the destination for large-scale logistics transportation, assign route numbers to the routes of each accessible route, and obtain the road condition time series data of each accessible route in the historical and current time periods;

[0007] Pre-training a road condition prediction model based on the road condition time series data of the historical time period, inputting the road condition time series data of the current time period into the road condition prediction model, obtaining the predicted road condition time series data of each reachable path within the future set time period, and selecting the optimal path based on the predicted road time series data;

[0008] When driving along the optimal route, the adjacent route segments of the next route are obtained, and the traffic volume of the next route is predicted based on the adjacent route segments and the traffic volume of the current route. If the traffic volume of the next route differs from the predicted traffic volume by more than a preset threshold, an alternative route for the next route is found.

[0009] Furthermore, all accessible routes from the departure point to the destination of large-scale logistics transportation are obtained, route numbers are assigned to the routes of each accessible route, and the road condition time series data of each accessible route in the historical and current time periods are obtained, including:

[0010] Mapping the logistics transportation network into a graph, wherein intersections, road entrances and exits in the logistics transportation network are regarded as nodes in the graph, and route segments in the logistics transportation network are represented as edges in the graph;

[0011] Obtain all accessible paths from the departure point to the destination in the logistics transportation network graph through the DFS method;

[0012] For each accessible path, check whether the restrictions on the path allow the length, width, height and weight characteristics of the large transport vehicle. If any edge restriction of a path is less than the corresponding parameter of the vehicle, the path is unreachable;

[0013] Obtain time series data on road conditions for each accessible route of large transport vehicles in historical and current time periods, obtain time series data on road conditions for historical and current time periods from road sensors, cameras, GPS tracking systems, vehicle electronic tags, and social media, and count data for different time periods, including peak hours, off-peak hours, holidays, and weekends, and the road condition time series data includes traffic volume and travel time.

[0014] Furthermore, a road condition prediction model is pre-trained based on historical time period data, and the road condition time series data of the current time period is input into the road condition prediction model to obtain the estimated road condition time series data and logistics-related factor time series data of each reachable path in the future set time period, including:

[0015] After cleaning and processing missing values ​​of the historical time period data, the data is divided into training set, test set and validation set;

[0016] Input historical road condition data into the time series analysis model for training, and obtain a prediction model after reaching a preset training cutoff condition;

[0017] Consider each edge of the reachable path of the large transport vehicle in the graph as a line, obtain the current traffic data of the first line of the reachable path from the starting point, input the current traffic data into the prediction model to obtain the first predicted travel time of the line, obtain the current traffic data of the second line, input the current traffic data into the prediction model to obtain the second predicted travel time of the line after the first predicted time, obtain the current traffic data of the third line, input the current traffic data into the prediction model to obtain the third predicted travel time of the line after the first plus the second predicted travel time, and so on, until the predicted travel time of the last line of the reachable path of the large transport vehicle is obtained;

[0018] The total loss of each reachable path is calculated using the loss model according to the predicted travel time.

[0019] Specifically, the loss model is:

[0020]

[0021] Among them, TL is the total loss of the reachable path, n is the number of lines on the reachable path, and f j is the historical fuel consumption of the transport vehicle passing through route j, l j is the length of line j, f c is the loss constant, is the predicted travel time through route j, t c is the time constant, m j is the cost of passing through route j, m c is the cost constant.

[0022] Furthermore, when driving along the optimal route, the adjacent route segment of the next route is obtained, and the traffic volume of the next route is predicted according to the traffic volume of the adjacent route segment and the current route. If the traffic volume of the next route differs from the predicted traffic volume by more than a preset threshold, an alternative route of the next route is found, including:

[0023] When walking along the optimal path, an adjacent route segment of the next route is obtained, where the adjacent route segment is an edge in the graph that is directly connected to an edge of the next route;

[0024] Predicting the traffic volume of the next route based on the traffic volume of the adjacent route segment and the current route;

[0025] If the traffic volume of the next route differs from the traffic volume predicted by deep learning by more than a preset threshold, an alternative route with the same departure and destination as the next route is searched in the logistics transportation network diagram;

[0026] The loss value of the replacement route is calculated through the loss model, and the route with the smallest sum of the loss and the weighted sum of the traffic volume calculated by the traffic flow dynamic evolution model is selected as the replacement driving route.

[0027] Further, predicting the traffic volume of the next route according to the traffic volume of the adjacent route segment and the current route includes:

[0028] The vehicle flow of the next line at time t is obtained through the basic flow model. Specifically, the basic flow model is:

[0029] F ij (t) = λ i (t)×w ij

[0030] Among them, F ij (t) is the traffic flow from route i to route j at time t, which quantifies the number of vehicles migrating from one specific area to another at a specific time point, and λ i(t) represents the vehicle generation rate of route i at time t, that is, the rate at which route i generates traffic flowing to other areas at that moment, w ij is the driving preference weight from route i to route j, indicating the preference of vehicles starting from i to j;

[0031] The traffic flow of the next line at time t+1 is predicted by the traffic flow dynamic evolution model. Specifically, the traffic flow dynamic evolution model is:

[0032]

[0033] Among them, f i (t+1) is the predicted traffic flow of line i at time point t+1, f i (t) is the observed traffic flow of line i at time point t, F ki (t) is the traffic flow from route k to route i, F ik (t) is the traffic flow from route i to route k, N i Represents the neighboring lines of line i.

[0034] The present invention proposes a road condition judgment system for large-scale logistics transportation, the system comprising:

[0035] The module for obtaining accessible paths and road condition data is used to obtain all accessible paths from the departure point to the destination for large-scale logistics transportation, assign a route number to each accessible path, and obtain the road condition time series data of each accessible path in the historical and current time periods;

[0036] The optimal path prediction module is used to pre-train a road condition prediction model based on the road condition time series data of the historical time period, input the road condition time series data of the current time period into the road condition prediction model, obtain the predicted road condition time series data of each reachable path within the future set time period, and select the optimal path according to the predicted road time series data;

[0037] The module for searching for replacement routes is used to obtain the adjacent route segments of the next route when traveling along the optimal route, predict the traffic volume of the next route based on the adjacent route segments and the traffic volume of the current route, and search for replacement routes for the next route if the traffic volume of the next route differs from the predicted traffic volume by more than a preset threshold.

[0038] Furthermore, the module for obtaining accessible paths and road condition data includes:

[0039] A mapping graph module, used for mapping the logistics transportation network into a graph, wherein the intersections, road entrances and exits in the logistics transportation network are regarded as nodes in the graph, and the route segments in the logistics transportation network are represented as edges in the graph;

[0040] Obtain all accessible paths module, which is used to obtain all accessible paths from the departure point to the destination in the logistics transportation network graph through the DFS method;

[0041] Get accessible path module. For each accessible path, check whether the restrictions on the path allow the length, width, height and weight characteristics of large transport vehicles. If any edge restriction of a path is less than the corresponding parameter of the vehicle, the path is unreachable.

[0042] The road condition data acquisition module is used to obtain the road condition time series data of each accessible route of large transport vehicles in the historical and current time periods, obtain the road condition time series data of the historical and current time periods from road sensors, cameras, GPS tracking systems, vehicle electronic tags and social media, and count the data of different time periods, including peak hours, non-peak hours, holidays and weekends. The road condition time series data includes traffic volume and travel time.

[0043] Furthermore, the optimal path prediction module includes:

[0044] The data processing module is used to clean the data of the historical time period and process the missing values, and then divide the data into training set, test set and validation set;

[0045] A training model module is used to input historical road condition data into a time series analysis model for training, and obtain a prediction model after reaching a preset training cutoff condition;

[0046] A prediction time window module is used to regard each edge of the reachable path of the large transport vehicle in the graph as a route, obtain the current traffic data of the first route of the reachable path from the starting point, input the current traffic data into the prediction model to obtain the first predicted travel time of the route, obtain the current traffic data of the second route, input the current traffic data into the prediction model to obtain the second predicted travel time of the route after the first predicted time, obtain the current traffic data of the third route, input the current traffic data into the prediction model to obtain the third predicted travel time of the route after the first plus the second predicted travel time, and so on, until the predicted travel time of the last route of the reachable path of the large transport vehicle is obtained;

[0047] A loss calculation module is used to calculate the total loss of each reachable path through a loss model according to the predicted travel time,

[0048] Specifically, the loss model is:

[0049]

[0050] Among them, TL is the total loss of the reachable path, n is the number of lines on the reachable path, and f jis the historical fuel consumption of the transport vehicle passing through route j, l j is the length of line j, f c is the loss constant, t j is the predicted travel time through route j, t c is the time constant, m j is the cost of passing through route j, m c is the cost constant.

[0051] Further, the search and replacement line module includes:

[0052] A module for obtaining adjacent route segments, used for obtaining adjacent route segments of the next route when walking along the optimal path, wherein the adjacent route segments are edges directly connected to the edges of the next route in the graph;

[0053] A vehicle flow prediction module, used for predicting the vehicle flow of the next route according to the vehicle flow of the adjacent route segment and the current route;

[0054] A comparison module, used to find an alternative route with the same departure and destination as the next route in the logistics transportation network diagram if the difference between the traffic volume of the next route and the traffic volume predicted by deep learning exceeds a preset threshold;

[0055] The module for obtaining a replacement driving route is used to calculate the loss value of the replacement route through a loss model, and select the route with the smallest sum of the loss and the weighted sum of the traffic volume calculated through the traffic flow dynamic evolution model as the replacement driving route.

[0056] Furthermore, the vehicle flow prediction module includes:

[0057] The vehicle flow acquisition module is used to obtain the vehicle flow of the next line at time t through the basic flow model. Specifically, the basic flow model is:

[0058] F ij (t) = λ i (t)×W ij

[0059] Among them, F ij (t) is the traffic flow from route i to route j at time t, which quantifies the number of vehicles migrating from one specific area to another at a specific time point, and λ i (t) represents the vehicle generation rate of route i at time t, that is, the rate at which route i generates traffic flowing to other areas at that moment, w ij is the driving preference weight from route i to route j, indicating the preference of vehicles starting from i to j;

[0060] The dynamic evolution traffic flow module is used to predict the traffic flow of the next line at time t+1 through the dynamic evolution model of traffic flow. Specifically, the dynamic evolution model of traffic flow is:

[0061]

[0062] Among them, f i (t+1) is the predicted traffic flow of line i at time point t+1, f i (t) is the observed traffic flow of line i at time point t, F ki (t) is the traffic flow from route k to route i, F ik (t) is the traffic flow from route i to route k, N i Represents the neighboring lines of line i.

[0063] The beneficial effects of the present invention are as follows: improving efficiency, optimizing route selection to reduce transportation time and ensure that goods can be delivered to the destination faster, which is particularly important for time-sensitive goods; reducing costs, by avoiding congestion and delays, reducing fuel consumption and possible waiting time costs, thereby reducing total transportation costs; enhancing adaptability, dynamic prediction and route adjustment functions make logistics transportation more adaptable, able to cope with sudden changes in road conditions and reduce unexpected impacts; improving service quality, more reliable transportation time predictions and improved service efficiency can greatly improve customer satisfaction and enhance the company's competitiveness; data-driven decision-making, data-based decision-making makes route planning more scientific and accurate, and reduces decision-making errors based on intuition or incomplete information; this technical solution greatly improves the efficiency and reliability of large-scale logistics transportation through the comprehensive application of path planning, real-time data monitoring, prediction models and dynamic adjustment strategies, and at the same time has the flexibility to cope with various changes in traffic conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 A schematic diagram of a method for determining road conditions for large-scale logistics transportation according to the present invention. DETAILED DESCRIPTION

[0065] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0066] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. The embodiments described are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0068] One embodiment of the present invention provides a method for determining road conditions for large-scale logistics transportation, the method comprising:

[0069] Obtain all accessible routes from the departure point to the destination for large-scale logistics transportation, assign route numbers to the routes of each accessible route, and obtain the road condition time series data of each accessible route in the historical and current time periods;

[0070] Pre-training a road condition prediction model based on the road condition time series data of the historical time period, inputting the road condition time series data of the current time period into the road condition prediction model, obtaining the predicted road condition time series data of each reachable path within the future set time period, and selecting the optimal path based on the predicted road time series data;

[0071] When driving along the optimal route, the adjacent route segments of the next route are obtained, and the traffic volume of the next route is predicted based on the adjacent route segments and the traffic volume of the current route. If the traffic volume of the next route differs from the predicted traffic volume by more than a preset threshold, an alternative route for the next route is found.

[0072] The working principle and effect of the above technical solution are as follows: path determination and number allocation, first obtain all feasible paths for large-scale logistics transportation from the departure point to the destination, which can be implemented on the traffic network diagram by using graph theory algorithms to determine all possible routes; each feasible path is assigned a unique route number for tracking and referencing each route in the system; data acquisition and model training, collect historical road condition time series data for the corresponding route on each reachable path, which may include traffic flow, vehicle speed, accident information and other traffic events; train a road condition prediction model based on these historical data, which can use various statistical or machine learning methods such as regression analysis, time series analysis or deep learning; real-time data processing and prediction, input the real-time road condition data collected in the current time period into the trained prediction model to predict the road conditions in the future; the prediction results help identify which paths may encounter delays or problems, so as to select the best path; dynamic route adjustment, during the journey, real-time monitoring and prediction of the traffic flow of the next route segment. If the predicted traffic volume exceeds the preset threshold, it indicates that there may be serious congestion or delays. In this case, the system will automatically find and recommend available alternative routes to ensure transportation efficiency and timeliness. Improve efficiency. Optimize route selection to reduce transportation time and ensure that goods can be delivered to the destination faster, which is particularly important for time-sensitive goods; reduce costs. By avoiding congestion and delays, reducing fuel consumption and possible waiting time costs, the total transportation cost can be reduced; enhance adaptability. Dynamic prediction and route adjustment functions make logistics transportation more adaptable and able to respond to sudden changes in road conditions and reduce unexpected impacts; improve service quality. More reliable transportation time predictions and improved service efficiency can greatly improve customer satisfaction and enhance the company's competitiveness; data-driven decision-making. Data-based decisions make route planning more scientific and accurate, reducing decision-making errors based on intuition or incomplete information; this technical solution greatly improves the efficiency and reliability of large-scale logistics transportation through the comprehensive application of path planning, real-time data monitoring, prediction models and dynamic adjustment strategies, while having the flexibility to respond to various changes in traffic conditions.

[0073] One embodiment of the present invention is a method for determining road conditions for large-scale logistics transportation, which obtains all accessible paths from a departure point to a destination for large-scale logistics transportation, assigns a path number to each accessible path, and obtains time series data of road conditions of each accessible path in a historical and current time period, including:

[0074] Mapping the logistics transportation network into a graph, wherein intersections, road entrances and exits in the logistics transportation network are regarded as nodes in the graph, and route segments in the logistics transportation network are represented as edges in the graph;

[0075] Obtain all accessible paths from the departure point to the destination in the logistics transportation network graph through the DFS method;

[0076] For each accessible path, check whether the restrictions on the path allow the length, width, height and weight characteristics of the large transport vehicle. If any edge restriction of a path is less than the corresponding parameter of the vehicle, the path is unreachable;

[0077] Obtain time series data on road conditions for each accessible route of large transport vehicles in historical and current time periods, obtain time series data on road conditions for historical and current time periods from road sensors, cameras, GPS tracking systems, vehicle electronic tags, and social media, and count data for different time periods, including peak hours, off-peak hours, holidays, and weekends, and the road condition time series data includes traffic volume and travel time.

[0078] The working principle and effect of the above technical solution are as follows: Graphical representation of logistics network, the logistics transportation network is converted into a graph, where nodes represent intersections, road entrances and exits, and edges represent road segments. This allows the transportation path to be explored and analyzed using graph algorithms; path exploration, using the depth-first search (DFS) method to explore all accessible paths from the starting point to the end point from the graph, DFS is a graph search algorithm suitable for this task that can fully explore all branches; path feasibility check, check whether each path is suitable for the physical size and weight restrictions of large transport vehicles. Any path that does not meet the vehicle size or weight requirements is marked as unreachable; road condition data acquisition and processing, collect historical and current road condition data for each route segment on the path, sources include road sensors, cameras, GPS tracking systems, vehicle electronic tags and social media; data is classified by time period (such as peak hours, off-peak hours, holidays, weekends), and key information such as traffic volume and travel time is extracted. Enhanced route selection optimization: by accurately mapping the logistics network and calculating all possible routes, the system can provide a series of feasible route options to optimize route selection; route feasibility checks ensure that all recommended routes comply with the restrictions of large transport vehicles, thereby avoiding possible regulatory issues and physical obstacles; improving transportation efficiency and reliability: based on historical and current road condition data, it can more accurately predict travel time and potential delays to ensure that transportation tasks are executed as planned; automated road condition monitoring and time series data analysis help transport vehicles avoid peak hours and congested areas, reducing travel time and costs; data-driven decision support: using real-time and historical data obtained from multiple data sources, making the decision-making process more scientific and accurate, and facilitating rapid response to sudden traffic incidents; accurate and reliable transportation services improve customer satisfaction and trust, and route selection and time management optimized by data analysis reduce the risk of cargo delays and improve service quality. This technical solution significantly improves the efficiency and reliability of large-scale logistics transportation and reduces transportation costs through a high degree of automation and precise data analysis.

[0079] One embodiment of the present invention is a method for judging road conditions for large-scale logistics transportation, which pre-trains a road condition prediction model based on historical time period data, inputs road condition time series data of the current time period into the road condition prediction model, and obtains estimated road condition time series data and logistics-related factor time series data of each reachable path within a future set time period, including:

[0080] After cleaning and processing missing values ​​of the historical time period data, the data is divided into training set, test set and validation set;

[0081] Input historical road condition data into the time series analysis model for training, and obtain a prediction model after reaching a preset training cutoff condition;

[0082] Consider each edge of the reachable path of the large transport vehicle in the graph as a line, obtain the current traffic data of the first line of the reachable path from the starting point, input the current traffic data into the prediction model to obtain the first predicted travel time of the line, obtain the current traffic data of the second line, input the current traffic data into the prediction model to obtain the second predicted travel time of the line after the first predicted time, obtain the current traffic data of the third line, input the current traffic data into the prediction model to obtain the third predicted travel time of the line after the first plus the second predicted travel time, and so on, until the predicted travel time of the last line of the reachable path of the large transport vehicle is obtained;

[0083] The total loss of each reachable path is calculated using the loss model according to the predicted travel time.

[0084] Specifically, the loss model is:

[0085]

[0086] Among them, TL is the total loss of the reachable path, n is the number of lines on the reachable path, and f j is the historical fuel consumption of the transport vehicle passing through route j, l j is the length of line j, f c is the loss constant, is the predicted travel time through route j, t c is the time constant, m j is the cost of passing through route j, m c is the cost constant.

[0087] The working principle and effect of the above technical solution are as follows: data cleaning and processing, preliminary cleaning of the collected road condition data of historical periods, including processing of missing values ​​and outliers to ensure data quality; dividing the cleaned data into training set, test set and validation set for modeling and verification of the effectiveness of the model; model training and prediction, using the time series analysis model to train the historical road condition data, and the model training will stop when the preset performance indicators (such as error minimization) are reached; using the trained model, inputting real-time or recently collected road condition data, and predicting the future travel time of each route segment on each path; continuous prediction recursive calculation, for each edge (route) on the accessible path of the large transport vehicle, from the starting point to the end point, the current road condition is obtained in turn, and the travel time is predicted, and the travel time from each route to the next route is predicted, and the total travel time of the entire route is calculated by accumulation; total loss calculation, using the loss model to calculate the total loss of the entire path, the model considers factors such as fuel consumption, travel time and cost, and the formula includes multiplying each factor by the attributes of the corresponding route, and then normalizing it with the corresponding constant. This method can quantify the total loss of different paths and help select the best path. Decision optimization: Through accurate predictions within the time window, accurate travel time predictions can be provided to help logistics companies optimize route selection, avoid possible delays, and reduce uncertainty in the transportation process; large-scale transport vehicles can plan routes more effectively, taking into account cost, time, and fuel consumption, thereby improving overall transportation efficiency; cost-effectiveness: through accurate loss calculations, transportation decisions can pay more attention to cost-effectiveness and choose the lowest-cost route; additional costs caused by fuel waste and overdue travel are reduced, thereby increasing overall profits; improved responsiveness: the system can respond quickly to immediate road conditions, update route planning in real time, reduce the impact caused by unpredictable factors such as traffic jams and road accidents, improve adaptability to emergencies, and ensure that goods arrive safely and on time; data-driven transparency: through data-driven methods, the decision-making process of logistics and transportation becomes more transparent and traceable, enhancing the trust of customers and stakeholders; providing quantifiable indicators to support decision-making, helping companies better explain and prove the logic behind the decision; this technical solution greatly improves the decision-making quality and execution efficiency of large-scale logistics and transportation through efficient data processing, accurate model prediction, and loss calculation that comprehensively considers multiple factors, reducing costs and improving service reliability.

[0088] An embodiment of the present invention is a method for determining road conditions for large-scale logistics transportation. When traveling along an optimal path, an adjacent route segment of a next route is obtained, and the traffic volume of the next route is predicted based on the traffic volume of the adjacent route segment and the current route. If the traffic volume of the next route differs from the predicted traffic volume by more than a preset threshold, a replacement route for the next route is found, including:

[0089] When walking along the optimal path, an adjacent route segment of the next route is obtained, where the adjacent route segment is an edge in the graph that is directly connected to an edge of the next route;

[0090] Predicting the traffic volume of the next route based on the traffic volume of the adjacent route segment and the current route;

[0091] If the traffic volume of the next route differs from the traffic volume predicted by deep learning by more than a preset threshold, an alternative route with the same departure and destination as the next route is searched in the logistics transportation network diagram;

[0092] The loss value of the replacement route is calculated through the loss model, and the route with the smallest sum of the loss and the weighted sum of the traffic volume calculated by the traffic flow dynamic evolution model is selected as the replacement driving route.

[0093] The working principle and effect of the above technical solution are: monitoring and predicting traffic flow. When a large transport vehicle travels along the optimal route, the system not only monitors the traffic flow of the current route, but also obtains the traffic flow information of the adjacent route segment directly connected to the next route segment. If the predicted traffic flow of the next route segment exceeds a certain preset threshold value and the traffic flow predicted by deep learning, it is considered that the route may be congested. In this case, the system will look for alternative routes in the logistics network diagram. These alternative routes need to have the same starting point and end point as the original route; compare the loss values ​​of all potential alternative routes and use the loss model for calculation. The model will consider factors such as travel time, fuel consumption and other related costs, weighted calculation of losses and traffic flow. In addition to calculating the above loss values, it will also consider the traffic flow calculated by the traffic dynamic evolution model; this integrated evaluation helps decision makers choose the route with the minimum comprehensive loss. Enhanced route optimization and management, dynamic monitoring and prediction of traffic flow, timely adjustment of transportation routes, avoid delays caused by potential traffic congestion, and reduce travel time and costs; optimize travel routes through scientific data-driven methods to improve transportation efficiency; cost-effectiveness is improved, and the use of deep learning and dynamic evolution model predictions can provide more accurate traffic flow information and reduce the additional costs caused by traffic jams; the application of loss models makes route selection not only based on time efficiency, but also takes into account economy and practicality; enhanced adaptability and flexibility, can respond quickly to sudden traffic changes, reduce the impact of unforeseen events by selecting alternative routes, and enhance the adaptability of large transport vehicles in complex and changeable traffic environments; clear route adjustments and traffic flow management make services more transparent and traceable. This technical solution effectively improves the efficiency and economy of large-scale logistics transportation through real-time traffic flow monitoring and prediction, dynamic route optimization and decision support, and comprehensive loss evaluation and management, while enhancing the controllability and reliability of the transportation process.

[0094] One embodiment of the present invention is a method for determining road conditions for large-scale logistics transportation, which predicts the traffic volume of the next route based on the traffic volume of the adjacent route segment and the current route, including:

[0095] The vehicle flow of the next line at time t is obtained through the basic flow model. Specifically, the basic flow model is:

[0096] F ij (t) = λ i (t)×w ij

[0097] Among them, F ij (t) is the traffic flow from route i to route j at time t, which quantifies the number of vehicles migrating from one specific area to another at a specific time point, and λ i (t) represents the vehicle generation rate of route i at time t, that is, the rate at which route i generates traffic flowing to other areas at that moment, w ij is the driving preference weight from route i to route j, indicating the preference of vehicles starting from i to j;

[0098] The traffic flow of the next line at time t+1 is predicted by the traffic flow dynamic evolution model. Specifically, the traffic flow dynamic evolution model is:

[0099]

[0100] Among them, f i (t+1) is the predicted traffic flow of line i at time point t+1, f i (t) is the observed traffic flow of line i at time point t, F ki (t) is the traffic flow from route k to route i, F ik (t) is the traffic flow from route i to route k, N i Represents the neighboring lines of line i.

[0101] The working principle and effect of the above technical solution are as follows: Application of the basic traffic model, which calculates the vehicle flow between two connected routes at a specific time t, and uses the vehicle generation rate λ of route i at time t i (t) and the driving preference weight w from route i to route j ijMultiplying, this reflects the amount that vehicles tend to migrate from one area to another; Implementation of the dynamic evolution model of traffic flow. According to the model, the predicted traffic flow of line i at the next time point t+1 is based on the traffic flow at the current time t plus the sum of the traffic flow flowing into line i, minus the sum of the traffic flow flowing out of line i. This takes into account the traffic flow dynamics of all adjacent lines of line i and predicts the flow direction and number of vehicles. Improve the accuracy of traffic prediction. By considering the traffic flow generation rate and driving preference weights between each line, this model can more accurately predict the traffic flow at a specific time and place, which is particularly effective for predicting traffic conditions during peak hours and in areas with high traffic jams; Optimize traffic management and regulation. Accurate traffic flow prediction enables urban traffic managers to formulate more effective traffic guidance and control strategies in advance, such as diverting traffic, optimizing traffic light control, and setting temporary traffic restrictions; Reduce traffic congestion. By real-time monitoring and predicting changes in traffic flow on different lines, traffic congestion can be effectively prevented and alleviated. For example, when it is predicted that a certain route will have high traffic volume, drivers can be notified in advance to choose other routes, or traffic diversion on that section can be strengthened; to improve logistics transportation efficiency, the logistics and transportation industry can use this accurate traffic flow prediction to optimize delivery routes and time plans, and reduce in-transit time and costs; this technical solution provides an accurate and efficient method to predict and manage traffic volume by comprehensively utilizing the basic traffic model of vehicle generation rate and driving preference weights, as well as the dynamic evolution model of traffic flow, thereby optimizing the overall operating efficiency and safety of the transportation system.

[0102] One embodiment of the present invention is a road condition determination system for large-scale logistics transportation, the system comprising:

[0103] The module for obtaining accessible paths and road condition data is used to obtain all accessible paths from the departure point to the destination for large-scale logistics transportation, assign a route number to each accessible path, and obtain the road condition time series data of each accessible path in the historical and current time periods;

[0104] The optimal path prediction module is used to pre-train a road condition prediction model based on the road condition time series data of the historical time period, input the road condition time series data of the current time period into the road condition prediction model, obtain the predicted road condition time series data of each reachable path within the future set time period, and select the optimal path according to the predicted road time series data;

[0105] The module for searching for replacement routes is used to obtain the adjacent route segments of the next route when traveling along the optimal route, predict the traffic volume of the next route based on the adjacent route segments and the traffic volume of the current route, and search for replacement routes for the next route if the traffic volume of the next route differs from the predicted traffic volume by more than a preset threshold.

[0106] The working principle and effect of the above technical solution are as follows: path determination and number allocation, first obtain all feasible paths for large-scale logistics transportation from the departure point to the destination, which can be implemented on the traffic network diagram by using graph theory algorithms to determine all possible routes; each feasible path is assigned a unique route number for tracking and referencing each route in the system; data acquisition and model training, collect historical road condition time series data for the corresponding route on each reachable path, which may include traffic flow, vehicle speed, accident information and other traffic events; train a road condition prediction model based on these historical data, which can use various statistical or machine learning methods such as regression analysis, time series analysis or deep learning; real-time data processing and prediction, input the real-time road condition data collected in the current time period into the trained prediction model to predict the road conditions in the future; the prediction results help identify which paths may encounter delays or problems, so as to select the best path; dynamic route adjustment, during the journey, real-time monitoring and prediction of the traffic flow of the next route segment. If the predicted traffic volume exceeds the preset threshold, it indicates that there may be serious congestion or delays. In this case, the system will automatically find and recommend available alternative routes to ensure transportation efficiency and timeliness. Improve efficiency. Optimize route selection to reduce transportation time and ensure that goods can be delivered to the destination faster, which is particularly important for time-sensitive goods; reduce costs. By avoiding congestion and delays, reducing fuel consumption and possible waiting time costs, the total transportation cost can be reduced; enhance adaptability. Dynamic prediction and route adjustment functions make logistics transportation more adaptable and able to respond to sudden changes in road conditions and reduce unexpected impacts; improve service quality. More reliable transportation time predictions and improved service efficiency can greatly improve customer satisfaction and enhance the company's competitiveness; data-driven decision-making. Data-based decisions make route planning more scientific and accurate, reducing decision-making errors based on intuition or incomplete information; this technical solution greatly improves the efficiency and reliability of large-scale logistics transportation through the comprehensive application of path planning, real-time data monitoring, prediction models and dynamic adjustment strategies, while having the flexibility to respond to various changes in traffic conditions.

[0107] In one embodiment of the present invention, the module for obtaining achievable paths and road condition data includes:

[0108] A mapping graph module, used for mapping the logistics transportation network into a graph, wherein the intersections, road entrances and exits in the logistics transportation network are regarded as nodes in the graph, and the route segments in the logistics transportation network are represented as edges in the graph;

[0109] Obtain all accessible paths module, which is used to obtain all accessible paths from the departure point to the destination in the logistics transportation network graph through the DFS method;

[0110] Get accessible path module. For each accessible path, check whether the restrictions on the path allow the length, width, height and weight characteristics of large transport vehicles. If any edge restriction of a path is less than the corresponding parameter of the vehicle, the path is unreachable.

[0111] The road condition data acquisition module is used to obtain the road condition time series data of each accessible route of large transport vehicles in the historical and current time periods, obtain the road condition time series data of the historical and current time periods from road sensors, cameras, GPS tracking systems, vehicle electronic tags and social media, and count the data of different time periods, including peak hours, non-peak hours, holidays and weekends. The road condition time series data includes traffic volume and travel time.

[0112] The working principle and effect of the above technical solution are as follows: Graphical representation of logistics network, the logistics transportation network is converted into a graph, where nodes represent intersections, road entrances and exits, and edges represent road segments. This allows the transportation path to be explored and analyzed using graph algorithms; path exploration, using the depth-first search (DFS) method to explore all accessible paths from the starting point to the end point from the graph, DFS is a graph search algorithm suitable for this task that can fully explore all branches; path feasibility check, check whether each path is suitable for the physical size and weight restrictions of large transport vehicles. Any path that does not meet the vehicle size or weight requirements is marked as unreachable; road condition data acquisition and processing, collect historical and current road condition data for each route segment on the path, sources include road sensors, cameras, GPS tracking systems, vehicle electronic tags and social media; data is classified by time period (such as peak hours, off-peak hours, holidays, weekends), and key information such as traffic volume and travel time is extracted. Enhanced route selection optimization: by accurately mapping the logistics network and calculating all possible routes, the system can provide a series of feasible route options to optimize route selection; route feasibility checks ensure that all recommended routes comply with the restrictions of large transport vehicles, thereby avoiding possible regulatory issues and physical obstacles; improving transportation efficiency and reliability: based on historical and current road condition data, it can more accurately predict travel time and potential delays to ensure that transportation tasks are executed as planned; automated road condition monitoring and time series data analysis help transport vehicles avoid peak hours and congested areas, reducing travel time and costs; data-driven decision support: using real-time and historical data obtained from multiple data sources, making the decision-making process more scientific and accurate, and facilitating rapid response to sudden traffic incidents; accurate and reliable transportation services improve customer satisfaction and trust, and route selection and time management optimized by data analysis reduce the risk of cargo delays and improve service quality. This technical solution significantly improves the efficiency and reliability of large-scale logistics transportation and reduces transportation costs through a high degree of automation and precise data analysis.

[0113] In one embodiment of the present invention, the optimal path prediction module includes:

[0114] The data processing module is used to clean the data of the historical time period and process the missing values, and then divide the data into training set, test set and validation set;

[0115] A training model module is used to input historical road condition data into a time series analysis model for training, and obtain a prediction model after reaching a preset training cutoff condition;

[0116] A prediction time window module is used to regard each edge of the reachable path of the large transport vehicle in the graph as a route, obtain the current traffic data of the first route of the reachable path from the starting point, input the current traffic data into the prediction model to obtain the first predicted travel time of the route, obtain the current traffic data of the second route, input the current traffic data into the prediction model to obtain the second predicted travel time of the route after the first predicted time, obtain the current traffic data of the third route, input the current traffic data into the prediction model to obtain the third predicted travel time of the route after the first plus the second predicted travel time, and so on, until the predicted travel time of the last route of the reachable path of the large transport vehicle is obtained;

[0117] A loss calculation module is used to calculate the total loss of each reachable path through a loss model according to the predicted travel time,

[0118] Specifically, the loss model is:

[0119]

[0120] Among them, TL is the total loss of the reachable path, n is the number of lines on the reachable path, and f j is the historical fuel consumption of the transport vehicle passing through route j, l j is the length of line j, f c is the loss constant, is the predicted travel time through route j, t c is the time constant, m j is the cost of passing through route j, m c is the cost constant. When the historical fuel consumption of route j (measured in fuel consumption per unit distance, such as liters / km) is multiplied by the actual length of route j (measured in kilometers), the calculated unit will be the unit of total fuel consumption, that is, liters, the loss constant value is 1, the unit is liters, the time constant value is 1, the unit is consistent with the predicted travel time, and the cost constant value is 1, the unit is consistent with mj.

[0121] The working principle and effect of the above technical solution are as follows: data cleaning and processing, preliminary cleaning of the collected road condition data of historical periods, including processing of missing values ​​and outliers to ensure data quality; dividing the cleaned data into training set, test set and validation set for modeling and verification of the effectiveness of the model; model training and prediction, using the time series analysis model to train the historical road condition data, and the model training will stop when the preset performance indicators (such as error minimization) are reached; using the trained model, inputting real-time or recently collected road condition data, and predicting the future travel time of each route segment on each path; continuous prediction recursive calculation, for each edge (route) on the accessible path of the large transport vehicle, from the starting point to the end point, the current road condition is obtained in turn, and the travel time is predicted, and the travel time from each route to the next route is predicted, and the total travel time of the entire route is calculated by accumulation; total loss calculation, using the loss model to calculate the total loss of the entire path, the model considers factors such as fuel consumption, travel time and cost, and the formula includes multiplying each factor by the attributes of the corresponding route, and then normalizing it with the corresponding constant. This method can quantify the total loss of different paths and help select the best path. Decision optimization: Through accurate predictions within the time window, accurate travel time predictions can be provided to help logistics companies optimize route selection, avoid possible delays, and reduce uncertainty in the transportation process; large-scale transport vehicles can plan routes more effectively, taking into account cost, time, and fuel consumption, thereby improving overall transportation efficiency; cost-effectiveness: through accurate loss calculations, transportation decisions can pay more attention to cost-effectiveness and choose the lowest-cost route; additional costs caused by fuel waste and overdue travel are reduced, thereby increasing overall profits; improved responsiveness: the system can respond quickly to immediate road conditions, update route planning in real time, reduce the impact caused by unpredictable factors such as traffic jams and road accidents, improve adaptability to emergencies, and ensure that goods arrive safely and on time; data-driven transparency: through data-driven methods, the decision-making process of logistics and transportation becomes more transparent and traceable, enhancing the trust of customers and stakeholders; providing quantifiable indicators to support decision-making, helping companies better explain and prove the logic behind the decision; this technical solution greatly improves the decision-making quality and execution efficiency of large-scale logistics and transportation through efficient data processing, accurate model prediction, and loss calculation that comprehensively considers multiple factors, reducing costs and improving service reliability.

[0122] In one embodiment of the present invention, the search and replacement line module includes:

[0123] A module for obtaining adjacent route segments, used for obtaining adjacent route segments of the next route when walking along the optimal path, wherein the adjacent route segments are edges directly connected to the edges of the next route in the graph;

[0124] A vehicle flow prediction module, used for predicting the vehicle flow of the next route according to the vehicle flow of the adjacent route segment and the current route;

[0125] A comparison module, used to find an alternative route with the same departure and destination as the next route in the logistics transportation network diagram if the difference between the traffic volume of the next route and the traffic volume predicted by deep learning exceeds a preset threshold;

[0126] The module for obtaining a replacement driving route is used to calculate the loss value of the replacement route through a loss model, and select the route with the smallest sum of the loss and the weighted sum of the traffic volume calculated through the traffic flow dynamic evolution model as the replacement driving route.

[0127] The working principle and effect of the above technical solution are: monitoring and predicting traffic flow. When a large transport vehicle travels along the optimal route, the system not only monitors the traffic flow of the current route, but also obtains the traffic flow information of the adjacent route segment directly connected to the next route segment. If the predicted traffic flow of the next route segment exceeds a certain preset threshold value and the traffic flow predicted by deep learning, it is considered that the route may be congested. In this case, the system will look for alternative routes in the logistics network diagram. These alternative routes need to have the same starting point and end point as the original route; compare the loss values ​​of all potential alternative routes and use the loss model for calculation. The model will consider factors such as travel time, fuel consumption and other related costs, weighted calculation of losses and traffic flow. In addition to calculating the above loss values, it will also consider the traffic flow calculated by the traffic dynamic evolution model; this integrated evaluation helps decision makers choose the route with the minimum comprehensive loss. Enhanced route optimization and management, dynamic monitoring and prediction of traffic flow, timely adjustment of transportation routes, avoid delays caused by potential traffic congestion, and reduce travel time and costs; optimize travel routes through scientific data-driven methods to improve transportation efficiency; cost-effectiveness is improved, and the use of deep learning and dynamic evolution model predictions can provide more accurate traffic flow information and reduce the additional costs caused by traffic jams; the application of loss models makes route selection not only based on time efficiency, but also takes into account economy and practicality; enhanced adaptability and flexibility, can respond quickly to sudden traffic changes, reduce the impact of unforeseen events by selecting alternative routes, and enhance the adaptability of large transport vehicles in complex and changeable traffic environments; clear route adjustments and traffic flow management make services more transparent and traceable. This technical solution effectively improves the efficiency and economy of large-scale logistics transportation through real-time traffic flow monitoring and prediction, dynamic route optimization and decision support, and comprehensive loss evaluation and management, while enhancing the controllability and reliability of the transportation process.

[0128] In one embodiment of the present invention, the vehicle flow prediction module includes:

[0129] The vehicle flow acquisition module is used to obtain the vehicle flow of the next line at time t through the basic flow model. Specifically, the basic flow model is:

[0130] F ij (t) = λ i (t)×w ij

[0131] Among them, F ij (t) is the traffic flow from route i to route j at time t, which quantifies the number of vehicles migrating from one specific area to another at a specific time point, and λ i (t) represents the vehicle generation rate of route i at time t, in vehicles / minute, that is, the rate at which route i generates traffic flows to other areas at that moment, w ij It is the driving preference weight from route i to route j, indicating the preference degree of vehicles departing from route i to route j. This can be obtained through statistics of historical time periods and can be expressed as the proportion of the traffic volume of destination route j in the historical time period to the total traffic volume departing from route i.

[0132] The dynamic evolution traffic flow module is used to predict the traffic flow of the next line at time t+1 through the dynamic evolution model of traffic flow. Specifically, the dynamic evolution model of traffic flow is:

[0133]

[0134] Among them, f i (t+1) is the predicted traffic flow of line i at time point t+1, f i (t) is the observed traffic flow of line i at time point t, F ki (t) is the traffic flow from route k to route i, F ik (t) is the traffic flow from route i to route k, N i Represents the neighboring lines of line i.

[0135] The working principle and effect of the above technical solution are as follows: Application of the basic traffic model, which calculates the vehicle flow between two connected routes at a specific time t, and uses the vehicle generation rate λ of route i at time t i (t) and the driving preference weight w from route i to route j ijMultiplying, this reflects the amount that vehicles tend to migrate from one area to another; Implementation of the dynamic evolution model of traffic flow. According to the model, the predicted traffic flow of line i at the next time point t+1 is based on the traffic flow at the current time t plus the sum of the traffic flow flowing into line i, minus the sum of the traffic flow flowing out of line i. This takes into account the traffic flow dynamics of all adjacent lines of line i and predicts the flow direction and number of vehicles. Improve the accuracy of traffic prediction. By considering the traffic flow generation rate and driving preference weights between each line, this model can more accurately predict the traffic flow at a specific time and place, which is particularly effective for predicting traffic conditions during peak hours and in areas with high traffic jams; Optimize traffic management and regulation. Accurate traffic flow prediction enables urban traffic managers to formulate more effective traffic guidance and control strategies in advance, such as diverting traffic, optimizing traffic light control, and setting temporary traffic restrictions; Reduce traffic congestion. By real-time monitoring and predicting changes in traffic flow on different lines, traffic congestion can be effectively prevented and alleviated. For example, when it is predicted that a certain route will have high traffic volume, drivers can be notified in advance to choose other routes, or traffic diversion on that section can be strengthened; to improve logistics transportation efficiency, the logistics and transportation industry can use this accurate traffic flow prediction to optimize delivery routes and time plans, and reduce in-transit time and costs; this technical solution provides an accurate and efficient method to predict and manage traffic volume by comprehensively utilizing the basic traffic model of vehicle generation rate and driving preference weights, as well as the dynamic evolution model of traffic flow, thereby optimizing the overall operating efficiency and safety of the transportation system.

[0136] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is intended to include these modifications and variations.

Claims

1. A method for judging road conditions for large-scale logistics transportation, characterized in that: The method comprises: Obtain all accessible routes from the departure point to the destination for large-scale logistics transportation, assign route numbers to the routes of each accessible route, and obtain the road condition time series data of each accessible route in the historical and current time periods; Pre-training a road condition prediction model based on the road condition time series data of the historical time period, inputting the road condition time series data of the current time period into the road condition prediction model, obtaining the predicted road condition time series data of each reachable path within the future set time period, and selecting the optimal path based on the predicted road time series data; When driving along the optimal route, obtain the adjacent route segments of the next route, and predict the traffic volume of the next route based on the adjacent route segments and the traffic volume of the current route. If the predicted traffic volume of the next route differs from the traffic volume predicted by deep learning by more than a preset threshold, find an alternative route for the next route, including: When walking along the optimal path, obtain the adjacent route segment of the next route, where the adjacent route segment is the edge in the graph that is directly connected to the edge of the next route: Predicting the traffic volume of the next route according to the traffic volume of the adjacent route segment and the current route includes: The vehicle flow of the next line at time t is obtained through the basic flow model. Specifically, the basic flow model is: F ij (t)=λ i (t)×w ij Among them, F ij (t) is the traffic flow from route i to route j at time t, which quantifies the number of vehicles migrating from one specific area to another at a specific time point, and λ i (t) represents the vehicle generation rate of route i at time t, that is, the rate at which route i generates traffic flowing to other areas at that moment, w ij is the driving preference weight from route i to route j, indicating the preference of vehicles starting from i to j; The traffic flow of the next line at time t+1 is predicted by the traffic flow dynamic evolution model. Specifically, the traffic flow dynamic evolution model is: Among them, fi(t+1) is the predicted traffic flow of line i at time point t+1, fi(t) is the observed traffic flow of line i at time point t, and F ki (t) is the traffic flow from route k to route i, F ik (t) is the traffic flow from route i to route k, N i represents the adjacent lines of line i; If the traffic volume of the next route differs from the traffic volume predicted by deep learning by more than a preset threshold, an alternative route with the same departure and destination as the next route is searched in the logistics transportation network diagram; The loss value of the replacement route is calculated through the loss model, and the route with the smallest sum of the loss and the weighted sum of the traffic volume calculated by the traffic flow dynamic evolution model is selected as the replacement driving route.

2. A method for determining road conditions for large-scale logistics transportation according to claim 1, characterized in that: Obtain all accessible routes from the departure point to the destination for large-scale logistics transportation, assign route numbers to the routes of each accessible route, and obtain the road condition time series data of each accessible route in the historical and current time periods, including: Mapping the logistics transportation network into a graph, wherein intersections, road entrances and exits in the logistics transportation network are regarded as nodes in the graph, and route segments in the logistics transportation network are represented as edges in the graph; Obtain all accessible paths from the departure point to the destination in the logistics transportation network graph through the DFS method; For each accessible path, check whether the restrictions on the path allow the length, width, height and weight characteristics of the large transport vehicle. If any edge restriction of a path is less than the corresponding parameter of the vehicle, the path is unreachable; Obtain time series data on road conditions for each accessible route of large transport vehicles in historical and current time periods, obtain time series data on road conditions for historical and current time periods from road sensors, cameras, GPS tracking systems, vehicle electronic tags, and social media, and count data for different time periods, including peak hours, off-peak hours, holidays, and weekends, and the road condition time series data includes traffic volume and travel time.

3. A method for determining road conditions for large-scale logistics transportation according to claim 1, characterized in that: The road condition prediction model is pre-trained based on the historical time period data, and the road condition time series data of the current time period is input into the road condition prediction model to obtain the estimated road condition time series data and logistics-related factor time series data of each reachable path in the future set time period, including: After cleaning and processing missing values ​​of the historical time period data, the data is divided into training set, test set and validation set; Input historical road condition data into the time series analysis model for training, and obtain a prediction model after reaching a preset training cutoff condition; Consider each edge of the reachable path of the large transport vehicle in the graph as a line, obtain the current traffic data of the first line of the reachable path from the starting point, input the current traffic data into the prediction model to obtain the first predicted travel time of the line, obtain the current traffic data of the second line, input the current traffic data into the prediction model to obtain the second predicted travel time of the line after the first predicted time, obtain the current traffic data of the third line, input the current traffic data into the prediction model to obtain the third predicted travel time of the line after the first plus the second predicted travel time, and so on, until the predicted travel time of the last line of the reachable path of the large transport vehicle is obtained; The total loss of each reachable path is calculated using the loss model according to the predicted travel time. Specifically, the loss model is: Among them, TL is the total loss of the reachable path, n is the number of lines on the reachable path, and f j is the historical fuel consumption of the transport vehicle passing through route j, l j is the length of line j, f c is the loss constant, is the predicted travel time through route j, t c is the time constant, m j is the cost of passing through route j, m c is the cost constant.

4. A road condition judgment system for large-scale logistics transportation, characterized in that: The system comprises: The module for obtaining accessible paths and road condition data is used to obtain all accessible paths from the departure point to the destination for large-scale logistics transportation, assign a route number to each accessible path, and obtain the road condition time series data of each accessible path in the historical and current time periods; The optimal path prediction module is used to pre-train a road condition prediction model based on the road condition time series data of the historical time period, input the road condition time series data of the current time period into the road condition prediction model, obtain the predicted road condition time series data of each reachable path within the future set time period, and select the optimal path according to the predicted road time series data; The module for searching for replacement routes is used to obtain the adjacent route segments of the next route when traveling along the optimal route, and predict the traffic volume of the next route based on the adjacent route segments and the traffic volume of the current route. If the traffic volume of the next route differs from the predicted traffic volume by more than a preset threshold, a replacement route for the next route is searched, including: A module for obtaining adjacent route segments, used for obtaining adjacent route segments of the next route when walking along the optimal path, wherein the adjacent route segments are edges directly connected to the edges of the next route in the graph; The vehicle flow prediction module is used to predict the vehicle flow of the next route according to the vehicle flow of the adjacent route segment and the current route, and includes: The vehicle flow acquisition module is used to obtain the vehicle flow of the next line at time t through the basic flow model. Specifically, the basic flow model is: F ij (t)=λ i (t)×w ij Among them, F ij (t) is the traffic flow from route i to route j at time t, which quantifies the number of vehicles migrating from one specific area to another at a specific time point, and λ i (t) represents the vehicle generation rate of route i at time t, that is, the rate at which route i generates traffic flowing to other areas at that moment, w ij is the driving preference weight from route i to route j, indicating the preference of vehicles starting from i to j; The dynamic evolution traffic flow module is used to predict the traffic flow of the next line at time t+1 through the dynamic evolution model of traffic flow. Specifically, the dynamic evolution model of traffic flow is: Among them, f i (t+1) is the predicted traffic flow of line i at time point t+1, f i (t) is the observed traffic flow of line i at time point t, F ki (t) is the traffic flow from route k to route i, F ik (t) is the traffic flow from route i to route k, N i represents the adjacent lines of line i; A comparison module, used to find an alternative route with the same departure and destination as the next route in the logistics transportation network diagram if the difference between the traffic volume of the next route and the traffic volume predicted by deep learning exceeds a preset threshold; The module for obtaining a replacement driving route is used to calculate the loss value of the replacement route through a loss model, and select the route with the smallest sum of the loss and the weighted sum of the traffic volume calculated through the traffic flow dynamic evolution model as the replacement driving route.

5. A road condition judgment system for large-scale logistics transportation according to claim 4, characterized in that: The module for obtaining accessible paths and road condition data includes: A mapping graph module, used for mapping the logistics transportation network into a graph, wherein the intersections, road entrances and exits in the logistics transportation network are regarded as nodes in the graph, and the route segments in the logistics transportation network are represented as edges in the graph; Obtain all accessible paths module, which is used to obtain all accessible paths from the departure point to the destination in the logistics transportation network graph through the DFS method; Get accessible path module. For each accessible path, check whether the restrictions on the path allow the length, width, height and weight characteristics of large transport vehicles. If any edge restriction of a path is less than the corresponding parameter of the vehicle, the path is unreachable. The road condition data acquisition module is used to obtain the road condition time series data of each accessible route of large transport vehicles in the historical and current time periods, obtain the road condition time series data of the historical and current time periods from road sensors, cameras, GPS tracking systems, vehicle electronic tags and social media, and count the data of different time periods, including peak hours, non-peak hours, holidays and weekends. The road condition time series data includes traffic volume and travel time.

6. A road condition judgment system for large-scale logistics transportation according to claim 4, characterized in that: The optimal path prediction module includes: The data processing module is used to clean the data of the historical time period and process the missing values, and then divide the data into training set, test set and validation set; A training model module is used to input historical road condition data into a time series analysis model for training, and obtain a prediction model after reaching a preset training cutoff condition; A prediction time window module is used to regard each edge of the reachable path of the large transport vehicle in the graph as a route, obtain the current traffic data of the first route of the reachable path from the starting point, input the current traffic data into the prediction model to obtain the first predicted travel time of the route, obtain the current traffic data of the second route, input the current traffic data into the prediction model to obtain the second predicted travel time of the route after the first predicted time, obtain the current traffic data of the third route, input the current traffic data into the prediction model to obtain the third predicted travel time of the route after the first plus the second predicted travel time, and so on, until the predicted travel time of the last route of the reachable path of the large transport vehicle is obtained; A loss calculation module is used to calculate the total loss of each reachable path through a loss model according to the predicted travel time, Specifically, the loss model is: Among them, TL is the total loss of the reachable path, n is the number of lines on the reachable path, and f j is the historical fuel consumption of the transport vehicle passing through route j, l j is the length of line j, f c is the loss constant, is the predicted travel time through route j, t c is the time constant, m j is the cost of passing through route j, m c is the cost constant.

Images